Neisseria meningitidis is a human pathogen that can cause acute meningitis and septicemia, with fatality rates around 15% [Girard et al, 2006]. Serogroup B meningitis accounts for 30-40% of meningitis cases in North America [Sharip et al, 2006; Kaplan et al, 2006] and up to 80% in some European countries [Trotter et al, 2007; Gray et al, 2006], yet a broadly protective vaccine is not available. Effective vaccines against other serogroups have been developed based on capsular polysaccharide conjugated to a carrier protein [Snape et al, 2008]. This approach was not feasible for serogroup B, due to poor immunogenicity [Morley et al, 2001] and concerns for vaccination-induced autoimmunity [Finne et al, 1983]. To date, vaccines based on outer membrane vesicles (OMV) are the only vaccines that successfully controlled serogroup B epidemics with examples in Norway, Cuba, and New Zealand [Bjune et al, 1991; Thornton et al, 2006; Martin et al, 1998; Sierra et al, Fredriksen et al, 1991].
OMV are released from the outer membrane of gram negative bacteria and consist of a phospholipid (PL) bilayer that contains outer membrane proteins, lipopolysacchharide (LPS) and periplasmic constituents [Deatherage et al, 2009]. PorA protein was identified as the major protective antigen in OMV, but is highly variable between the circulating serogroup B strains which complicates vaccine development [Saukkonen et al, 1989; Martin et al, 2006]. For this reason, Rijks Instituut voor Volksgezondheid en Milieu (RWM), i.e the National Institute for Public Health and the Environment (Bilthoven, The Netherlands) developed an OMV vaccine based on genetically modified N. meningitidis strains that express multiple PorA subtypes. This multivalent OMV vaccine was initially made with 2 trivalent PorA strains, expressing a total of 6 PorA subtypes [van der Ley et al, 1995; Claassen et al, 1996] and provided functional immunogenicity in phase II clinical trials. To ensure sufficient coverage for serogroup B strains circulating globally, a third trivalent strain was added [van den Dobbelsteen et al, 2007].
OMV vaccines are traditionally prepared with detergent extraction (dOMV purification process) to remove LPS and increase vesicle release. The LPS of N. meningitidis is highly toxic, but residual amounts (approx. 1%) are needed to maintain vesicle structure and adjuvate the immune response against PorA [Arigita et al, 2005; Arigita et al, 2003; Steeghs et al, 2004]. With balanced detergent concentrations the dOMV purification process provides these requirements, however there are major disadvantages. Along with LPS, detergent removes PL and also lipoproteins that contribute to immunogenicity, such as factor H binding protein [Koeberling et al, 2009; Koeberling et al, 2008]. The resulting immune response is directed against a specific PorA subtype, without eliciting cross-protection [Morley et al, 2001; van der Voort et al, 1996]. In addition, selective removal of LPS and PL changes the native vesicle structure and promotes aggregation [Holst et al, 2009; Cametti et al, 2008]. Detergent-treatment is necessary to decrease LPS toxicity, but has detrimental side effects that complicate vaccine development.
Detergent-free OMV purification processes retain all LPS, resulting not only in a preserved native vesicle structure, but also in vaccines that are inherently toxic when used for parenteral immunization [Holst et al, 2009]. Two detergent-free purification processes have been described. The native OMV (nOMV) process [Zollinger et al 1979; U.S. Pat. No. 6,558,677] comprises similar steps as dOMV, however with a detergent-free extraction step and the supernatant OMV (sOMV) process [Post et al, 2005; Devoe et al, 1973; Hoekstra et al, 1976] utilizes ultrafiltration or ultracentrifugation to purify spontaneously released OMV from the culture supernatant, without extraction. nOMV vaccines produced encouraging results in animals and humans, but high LPS content limited applicability of the vaccine to intranasal administration [Guthrie et al, 2004; Katial et al, 2002; Saunders et al, 1999; Drabick et al, 1999]. Preclinical data on sOMV vaccines is limited to a single study in mice, reporting cross-protection against a panel of serogroup B strains that was not found with dOMV, however potential differences in toxicity and stability were not addressed [Ferrari et al, 2006]. The sOMV process imposes an additional challenge, since it produces OMV yields that are too low for feasible process development [Post et al, 2005; Devoe et al, 1973].
Discovery of lpxL1 mutant strains at RIVM Bilthoven [van der Ley et al, 2001] provided a solution for the LPS toxicity issue. Deletion of lpxL1 attenuate LPS toxicity, while preserving the adjuvant activity needed for the immune response [Koeberling et al, 2008; van der Ley et al, 2001; Fisseha et al, 2005; van de Waterbeemd et al, 2010].
However, for clinical trials or GMP manufacturing of nOMV/sOMV a robust scalable production process is required in which the EDTA extraction step is needed for high yield, but to date causes undesired effects such as bacterial lysis [Prachayasittikul et al, 2010]. DNA release caused by lysis is a problem for large scale production since removal processes such as ultracentrifugation only have limited capacity.
Accordingly, since the processes available to date to prepare sOMV and nOMV either suffer from low yield and/or low purity and/or are limited to laboratory scale, there is a need for improved processes to prepare bacterial OMV, in particular for processes at industrial scale.